The Application of Neural Networks in Detecting Spina Bifida on Prenatal Ultrasound Images

Detection of Spina Bifida Pathology in Prenatal Ultrasound Images is a critical task in perinatal medicine. Existing ultrasound diagnostic methods heavily rely on the skills of the physician and the quality of the equipment, which can lead to errors in image interpretation. This paper proposes an approach that uses a combination of deep learning models for the automatic analysis of ultrasound images. The process consists of two stages: first, an object detection model is used to localize the region of interest in the image. Then, using an ensemble of neural networks, the images are classified as suitable or unsuitable for diagnosis, and the presence or absence of pathology is determined. The research results show high efficiency of the proposed method, which opens prospects for its implementation in clinical practice.